Annotation of parser3/src/include/pa_string.h, revision 1.144.2.11

1.41      paf         1: /** @file
1.43      paf         2:        Parser: string class decl.
                      3: 
1.144     paf         4:        Copyright (c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com)
1.124     paf         5:        Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.1       paf         6: */
                      7: 
                      8: #ifndef PA_STRING_H
                      9: #define PA_STRING_H
1.140     paf        10: 
1.144.2.11! paf        11: static const char* IDENT_STRING_H="$Date: 2003/01/28 12:11:50 $";
1.144.2.10  paf        12: 
                     13: // includes
1.1       paf        14: 
1.15      paf        15: #include "pa_pool.h"
1.4       paf        16: 
1.144.2.5  paf        17: // forwards
                     18: 
                     19: class String;
                     20: 
                     21: typedef object_ptr<String> StringPtr;
                     22: typedef object_ptr<const String> ConstStringPtr;
                     23: const ConstStringPtr ConstStringPtrZero(0);
                     24: 
1.144.2.10  paf        25: // includes
                     26: 
                     27: #include "pa_types.h"
                     28: #include "pa_array.h"
                     29: 
                     30: // forwards
                     31: 
1.144.2.9  paf        32: class Table;
                     33: typedef object_ptr<Table> TablePtr;
1.144.2.5  paf        34: class SQL_Connection;
                     35: class Dictionary;
                     36: 
                     37: // helpers
                     38: 
1.9       paf        39: #ifndef NO_STRING_ORIGIN
1.33      paf        40: #      define STRING_APPEND_PARAMS \
                     41:                const char *src, size_t size,  \
1.144.2.7  paf        42:                String_UL lang, \
1.33      paf        43:                const char *file, uint line
1.54      paf        44: /// appends piece to String  @see String::real_append
                     45: #      define APPEND(src, size, lang, file, line) \
                     46:                real_append(src, size, lang, file, line)
1.9       paf        47: #else
1.33      paf        48: #      define STRING_APPEND_PARAMS \
                     49:                const char *src, \
                     50:                size_t size, \
1.144.2.7  paf        51:                String_UL lang
1.54      paf        52: /// appends piece to String  @see String::real_append
                     53: #      define APPEND(src, size, lang, file, line) \
                     54:                real_append(src, size, lang)
                     55: #endif
1.83      paf        56: /// appends clean piece to String @see String::real_append
1.54      paf        57: #define APPEND_CLEAN(src, size, file, line) \
1.62      paf        58:        APPEND(src, size, String::UL_CLEAN, file, line)
1.83      paf        59: /// appends piece to String as-is @see String::real_append
                     60: #define APPEND_AS_IS(src, size, file, line) \
                     61:        APPEND(src, size, String::UL_AS_IS, file, line)
1.48      paf        62: /// appends tainted piece to String  @see String::real_append
1.54      paf        63: #define APPEND_TAINTED(src, size, file, line) \
1.62      paf        64:        APPEND(src, size, String::UL_TAINTED, file, line)
1.48      paf        65: /// handy: appends const char* piece to String  @see String::real_append
1.83      paf        66: #define        APPEND_CONST(src) APPEND_AS_IS(src, 0, 0, 0)
1.9       paf        67: 
1.144.2.7  paf        68: typedef uint String_UL;
                     69: 
1.144.2.2  paf        70: struct String_fragment { 
1.144.2.7  paf        71:        String_UL/*Untaint_lang*/ lang; ///< untaint flag, later untaint language
1.144.2.2  paf        72:        const char *ptr;  ///< pointer to the start
                     73:        size_t size;  ///< length
1.41      paf        74: 
1.144.2.2  paf        75: #ifndef NO_STRING_ORIGIN
                     76:        /// all String pieces hold information of where they come from
                     77:        struct Origin {
                     78:                const char *file;  ///< macros file name | load file name | sql query text
                     79:                uint line; ///< file line no | record no
                     80:        } origin;  ///< origin
                     81: #endif
                     82: };
                     83: 
                     84: /** 
                     85:        String which knows the language of all it's fragments.
1.41      paf        86: 
                     87:        All pieces remember 
                     88:        - the file and its line they are from [can be turned off by NO_STRING_ORIGIN]
                     89:        - whether they are tainted or not, 
                     90:          and the language which should be used to detaint them
                     91: */
1.48      paf        92: 
1.144.2.2  paf        93: class String: public Array<String_fragment> {
                     94:        size_t fsize;
1.144.2.5  paf        95:        int count() const; // hiding fro accidental use instead of size()
1.144.2.2  paf        96: public:
1.1       paf        97: 
1.106     parser     98:        /** piece is tainted or not. the language to use when detaint
                     99:                remember to change String_Untaint_lang_name @ untaint.C along
                    100:        */
1.27      paf       101:        enum Untaint_lang {
1.72      paf       102:                UL_UNSPECIFIED=0, ///< zero value handy for hash lookup @see untaint_lang_name2enum
1.122     paf       103:                // these two must go before others, there are checks for >UL_AS_IS
1.62      paf       104:                UL_CLEAN, ///< clean
1.122     paf       105:                UL_AS_IS,     ///< leave all characters intact
                    106: 
1.48      paf       107:                UL_PASS_APPENDED,
1.41      paf       108:                        /**<
                    109:                                leave language built into string being appended.
                    110:                                just a flag, that value not stored
                    111:                        */
1.122     paf       112:                UL_TAINTED,  ///< tainted, untaint language as assigned later 
                    113:                // untaint languages. assigned by ^untaint[lang]{...}
1.103     parser    114:                UL_FILE_SPEC, ///< file specification
1.74      paf       115:                UL_HTTP_HEADER,    ///< text in HTTP response header
                    116:                UL_MAIL_HEADER,    ///< text in mail header
1.48      paf       117:                UL_URI,       ///< text in uri
                    118:                UL_TABLE,     ///< ^table:set body
                    119:                UL_SQL,       ///< ^table:sql body
                    120:                UL_JS,        ///< JavaScript code
1.102     parser    121:                UL_XML,         ///< ^dom:set xml
1.48      paf       122:                UL_HTML,      ///< HTML code (for editing)
1.144.2.7  paf       123:                UL_OPTIMIZE_BIT = 0x8000  ///< flag, requiring cstr whitespace optimization
1.27      paf       124:        };
                    125: 
1.8       paf       126: public:
                    127: 
1.144.2.1  paf       128:        //static String& OnPool(Pool& apool, const char *local_src=0, size_t src_size=0, bool tainted=false);
1.144.2.6  paf       129:        explicit String(const char *src=0, size_t src_size=0, bool tainted=false);
1.14      paf       130:        String(const String& src);
1.144.2.2  paf       131:        bool is_empty() const { return !size(); }
                    132:        size_t size() const { return fsize; }
1.69      paf       133:        /// convert to C string. if 'lang' known, forcing 'lang' to it
1.144.2.5  paf       134:        CharPtr cstr(Untaint_lang lang=UL_AS_IS, 
1.82      paf       135:                SQL_Connection *connection=0,
1.144.2.11! paf       136:                const char *charset_name=0) const {
1.88      parser    137: 
1.144.2.11! paf       138:                CharPtr result(new char[cstr_bufsize(lang, connection, charset_name)]);
        !           139:                char *eol=store_to(result.get(), lang, connection, charset_name);
1.50      paf       140:                *eol=0;
                    141:                return result;
                    142:        }
1.144.2.9  paf       143:        char *cstr(Pool& pool, Untaint_lang lang=UL_AS_IS, 
                    144:                SQL_Connection *connection=0,
1.144.2.11! paf       145:                const char *charset_name=0) const {
1.144.2.9  paf       146: 
1.144.2.11! paf       147:                char *result=new(pool) char[cstr_bufsize(lang, connection, charset_name)];
        !           148:                char *eol=store_to(result, lang, connection, charset_name);
1.144.2.9  paf       149:                *eol=0;
                    150:                return result;
                    151:        }
1.117     paf       152:        char *cstr_debug_origins() const;
1.108     parser    153:        /// puts pieces to buf
1.144.2.5  paf       154:        void serialize(Pool& pool, size_t prolog_size,  char *& buf, size_t& buf_size) const;
1.108     parser    155:        /// appends pieces from buf to self
1.132     paf       156:        bool deserialize(size_t prolog_size, void *buf, size_t buf_size, const char *file);
1.46      paf       157:        /** append fragment
1.83      paf       158:                @see APPEND_AS_IS, APPEND_CLEAN, APPEND_TAINTED, APPEND_CONST
1.46      paf       159:        */
1.9       paf       160:        String& real_append(STRING_APPEND_PARAMS);
1.44      paf       161:        /// @return <0 ==0 or >0 depending on comparison result
1.62      paf       162:        int cmp (int& partial, const String& src, 
1.72      paf       163:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.56      paf       164:        bool operator < (const String& src) const {     int p; return cmp(p, src)<0; }
                    165:        bool operator > (const String& src) const {     int p; return cmp(p, src)>0; }
                    166:        bool operator <= (const String& src) const { int p; return cmp(p, src)<=0; }
                    167:        bool operator >= (const String& src) const { int p; return cmp(p, src)>=0; }
1.26      paf       168:        bool operator == (const String& src) const {
                    169:                if(size()!=src.size()) // can speed up in trivial case
                    170:                        return false;
1.56      paf       171:                int p; return cmp(p, src)==0;
1.26      paf       172:        }
1.56      paf       173:        bool operator != (const String& src) const { int p; return cmp(p, src)!=0; }
1.26      paf       174: 
1.50      paf       175:        /**
                    176:                 @param partial 
                    177:                        returns partial match status. 
1.51      paf       178:                        - -1: strings too different
                    179:                        -  0: full match
                    180:                        -  1: means @c this starts @c src
                    181:                        -  2: means @src starts @this
1.50      paf       182:        */
1.61      paf       183:        int cmp(int& partial, const char* src_ptr, size_t src_size=0, 
1.72      paf       184:                size_t this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.139     paf       185:        /// this starts with src
                    186:        bool starts_with(const char* src_ptr, size_t src_size=0) const {
                    187:                int p; cmp(p, src_ptr, src_size);
                    188:                return p==0 || p==2;
                    189:        }
1.50      paf       190:        bool operator == (const char* src_ptr) const { 
                    191:                size_t src_size=src_ptr?strlen(src_ptr):0;
                    192:                if(size() != src_size)
                    193:                        return false;
                    194:                int partial; // unused
1.56      paf       195:                return cmp(partial, src_ptr, src_size)==0; 
1.50      paf       196:        }
1.80      paf       197:        bool operator != (const char* src_ptr) const { 
                    198:                int partial; // unused
                    199:                return cmp(partial, src_ptr, 0)!=0; 
                    200:        }
                    201: 
1.144.2.11! paf       202:        String& append_to(String& dest, uchar lang, bool forced) const;
        !           203:        String& append(const String& src, uchar lang, bool forced=false) { 
        !           204:                return src.append_to(*this, lang, forced);
        !           205:        }
1.76      paf       206:        String& operator << (const String& src) { return append(src, UL_PASS_APPENDED); }
1.144.2.8  paf       207:        String& operator << (const StringPtr src) { return append(*src, UL_PASS_APPENDED); }
1.76      paf       208:        String& operator << (const char *src) { return APPEND_CONST(src); }
1.8       paf       209: 
1.144.2.7  paf       210:        /// simple hash code of string
1.14      paf       211:        uint hash_code() const;
1.100     parser    212: 
1.142     paf       213:        /// extracts first char of a string, if any
                    214:        char first_char() const {
1.144.2.2  paf       215:                return is_empty()?0:*felements[0].ptr;
1.142     paf       216:        }
1.54      paf       217: 
                    218:        /// extracts [start, finish) piece of string
1.144.2.4  paf       219:        StringPtr mid(size_t start, size_t finish) const;
1.55      paf       220: 
1.59      paf       221:        /// @return position of substr in string, -1 means "not found" [String version]
1.62      paf       222:        int pos(const String& substr, 
1.112     paf       223:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.59      paf       224:        /// @return position of substr in string, -1 means "not found" [const char* version]
1.97      parser    225:        int pos(const char *substr, size_t substr_size=0, 
1.112     paf       226:                int this_offset=0, Untaint_lang lang=UL_UNSPECIFIED) const;
1.62      paf       227: 
1.144.2.5  paf       228:        void split(Array<ConstStringPtr>& result, 
1.62      paf       229:                size_t *pos_after_ref, 
                    230:                const char *delim, size_t delim_size, 
1.86      paf       231:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.144.2.5  paf       232:        void split(Array<ConstStringPtr>& result, 
1.62      paf       233:                size_t *pos_after_ref, 
                    234:                const String& delim, 
1.86      paf       235:                Untaint_lang lang=UL_UNSPECIFIED, int limit=-1) const;
1.62      paf       236: 
1.144.2.9  paf       237:        typedef void (*Row_action)(Table& table, Array<ConstStringPtr> *row, 
1.136     paf       238:                int prestart, int prefinish, 
                    239:                int poststart, int postfinish,
1.144.2.9  paf       240:                void *info);
1.87      parser    241:        /**
1.144.2.5  paf       242:                @return table of found items, if any.
1.87      parser    243:                table format is defined and fixed[can be used by others]: 
                    244:                @verbatim
                    245:                        prematch/match/postmatch/1/2/3/...
                    246:                @endverbatim
                    247:        */
1.144.2.9  paf       248:        TablePtr match(Charset& source_charset,
1.144.2.5  paf       249:                ConstStringPtr aorigin,         
1.64      paf       250:                const String& regexp, 
1.144.2.5  paf       251:                ConstStringPtr options,
1.99      parser    252:                Row_action row_action, void *info,
1.144.2.9  paf       253:                bool *was_global=0) const;
1.87      parser    254:        enum Change_case_kind {
                    255:                CC_UPPER,
                    256:                CC_LOWER
                    257:        };
1.144.2.5  paf       258:        StringPtr change_case(Pool& pool, Charset& source_charset,
1.87      parser    259:                Change_case_kind kind) const;
1.144.2.5  paf       260:        StringPtr replace(Pool& pool, Dictionary& dict) const;
1.96      parser    261:        double as_double() const;
                    262:        int as_int() const;
1.137     paf       263: 
1.144.2.5  paf       264:        StringPtr join_chains(Pool& pool) const;
1.13      paf       265: 
1.41      paf       266: #ifndef NO_STRING_ORIGIN
                    267:        /// origin of string. calculated by first row
1.144.2.2  paf       268:        const String_fragment::Origin& origin() const;
1.41      paf       269: #endif
1.8       paf       270: 
1.144.2.5  paf       271:        size_t cstr_bufsize(Untaint_lang lang=UL_UNSPECIFIED,
                    272:                SQL_Connection *connection=0,
1.144.2.11! paf       273:                const char *charset_name=0) const;
1.79      paf       274:        /// convert to C string, store to 'dest' which must be big enough for proper untaint
1.82      paf       275:        char *store_to(char *dest, Untaint_lang lang=UL_UNSPECIFIED, 
                    276:                SQL_Connection *connection=0,
1.144.2.11! paf       277:                const char *charset_name=0) const;
1.144.2.2  paf       278: /*
1.144.2.1  paf       279:        void join_chain(
1.130     paf       280:                                           const Chunk*& achunk, const Chunk::Row*& arow, uint& acountdown, 
1.119     paf       281:                                           uchar& joined_lang, const char *& joined_ptr, size_t& joined_size) const;
1.144.2.2  paf       282: */
1.7       paf       283: private: //disabled
                    284: 
1.12      paf       285:        String& operator = (const String&) { return *this; }
1.7       paf       286: 
1.1       paf       287: };
1.144.2.7  paf       288: 
                    289: /// simple hash code of string. used by Hash
                    290: inline uint hash_code(const String& self) {
                    291:        return self.hash_code();
                    292: }
                    293: /// simple hash code of string. used by Hash
                    294: inline uint hash_code(ConstStringPtr self) {
                    295:        return self->hash_code();
                    296: }
1.144.2.5  paf       297: 
                    298: // helpers
                    299: 
                    300: #define STRING_PREPARED_FOREACH_FRAGMENT(code) \
                    301:        for(; fragment<last; fragment++) { \
                    302:                code; \
                    303:        }
                    304: 
                    305: #define STRING_FOREACH_FRAGMENT(code) \
                    306:        { \
                    307:                element_type *last = felements+fused; \
                    308:                element_type *fragment = felements; \
                    309:                STRING_PREPARED_FOREACH_FRAGMENT(code) \
                    310:        }
1.1       paf       311: 
                    312: #endif

E-mail: